The Department of Defense has formally notified Anthropic that its Claude models are a supply‑chain risk, a designation that can restrict use of the company’s products by defense contractors and federal systems. The Pentagon said the move reflected a need for the military to use technology for “all lawful purposes.” Anthropic announced it will challenge the decision in court and said the designation is narrower than officials initially described. Anthropic CEO Dario Amodei also sought to walk back a leaked internal memo that criticized competitors and employees at other AI firms. The dispute has already prompted some defense contractors to reexamine vendor ties and spotlights the political and legal friction between AI firms that set product restrictions and a Pentagon pressing for unrestricted operational use.